基于音频特征MFCC的混合域脆弱水印算法  

Fragile watermarking algorithm in hybrid domain based on audio features of MFCC

作　　者：宋慧[1,2] 李晨[1] 田丽华[1] 张玉龙[1] 

机构地区：[1]西安交通大学软件学院,陕西西安710049 [2]武警工程大学信息工程系,陕西西安710086

出　　处：《计算机工程与设计》2017年第7期1885-1890,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61403302);西安交通大学科研业务基金项目(XJJ2016029)

摘　　要：将DWT与SVD进行结合,设计一种以音频自身特征MFCC为基础的脆弱水印方案。不直接采用二值图像作为水印,利用原始音频特征Mel频率倒谱系数(MFCC)对常见音频处理操作比较敏感的特性,一方面构造脆弱水印,另一方面进行音频内容的认证。利用音频特征MFCC进行脆弱水印构造,对各音频分段进行DWT变换,选取小波域的低频部分并进行SVD分解操作,将水印通过均值量化的策略嵌入在DWT变换和SVD分解组成的混合域中。实验结果表明,该方案有很好的透明性,对多种攻击具备较强的敏感性,可对恶意篡改进行定位。Combining discrete wavelet transform with singular value decomposition,a fragile watermarking algorithm based on audio features of MFCC was presented.The algorithm did not adopt binary image as the watermark,but utilized the features of the original audio Mel frequency cepstral coefficients（MFCC） to construct the fragile watermark sequence,since MFCC was very sensitive to common audio attacks,and this feature was also applied to audio content authentication.The audio features were used to construct the fragile watermark sequence.Each audio segment was processed through discrete wavelet and the low frequency parts of wavelet transform were selected for singular value decomposition.The watermark sequence was embedded in hybrid domain using mean quantization method.Experimental results indicate that the method presented is invisible and sensitive enough to the popular attacks,and it can locate the malicious tampering.

关 键 词：脆弱水印 音频水印 奇异值分解 量化 篡改定位 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]